|
a |
|
b/ipynb/classifier.ipynb |
|
|
1 |
{ |
|
|
2 |
"cells": [ |
|
|
3 |
{ |
|
|
4 |
"cell_type": "code", |
|
|
5 |
"execution_count": 1, |
|
|
6 |
"metadata": {}, |
|
|
7 |
"outputs": [ |
|
|
8 |
{ |
|
|
9 |
"data": { |
|
|
10 |
"text/plain": [ |
|
|
11 |
"True" |
|
|
12 |
] |
|
|
13 |
}, |
|
|
14 |
"execution_count": 1, |
|
|
15 |
"metadata": {}, |
|
|
16 |
"output_type": "execute_result" |
|
|
17 |
} |
|
|
18 |
], |
|
|
19 |
"source": [ |
|
|
20 |
"import numpy as np\n", |
|
|
21 |
"import pandas as pd\n", |
|
|
22 |
"from sklearn.model_selection import train_test_split\n", |
|
|
23 |
"from fastai import *\n", |
|
|
24 |
"from fastai.basic_data import *\n", |
|
|
25 |
"from fastai.basic_train import *\n", |
|
|
26 |
"from fastai.tabular import *\n", |
|
|
27 |
"from torch import nn\n", |
|
|
28 |
"import torch.nn.functional as F\n", |
|
|
29 |
"from torch.utils.data import Dataset\n", |
|
|
30 |
"\n", |
|
|
31 |
"torch.cuda.is_available()" |
|
|
32 |
] |
|
|
33 |
}, |
|
|
34 |
{ |
|
|
35 |
"cell_type": "markdown", |
|
|
36 |
"metadata": {}, |
|
|
37 |
"source": [ |
|
|
38 |
"### Load ksent vectors of random genome samples of 16kb" |
|
|
39 |
] |
|
|
40 |
}, |
|
|
41 |
{ |
|
|
42 |
"cell_type": "code", |
|
|
43 |
"execution_count": null, |
|
|
44 |
"metadata": {}, |
|
|
45 |
"outputs": [], |
|
|
46 |
"source": [ |
|
|
47 |
"df = pd.read_pickle(\"../data/random-ksent.pkl\")" |
|
|
48 |
] |
|
|
49 |
}, |
|
|
50 |
{ |
|
|
51 |
"cell_type": "markdown", |
|
|
52 |
"metadata": {}, |
|
|
53 |
"source": [ |
|
|
54 |
"#### Look at the data" |
|
|
55 |
] |
|
|
56 |
}, |
|
|
57 |
{ |
|
|
58 |
"cell_type": "code", |
|
|
59 |
"execution_count": 2, |
|
|
60 |
"metadata": { |
|
|
61 |
"scrolled": false |
|
|
62 |
}, |
|
|
63 |
"outputs": [ |
|
|
64 |
{ |
|
|
65 |
"data": { |
|
|
66 |
"text/html": [ |
|
|
67 |
"<div>\n", |
|
|
68 |
"<style scoped>\n", |
|
|
69 |
" .dataframe tbody tr th:only-of-type {\n", |
|
|
70 |
" vertical-align: middle;\n", |
|
|
71 |
" }\n", |
|
|
72 |
"\n", |
|
|
73 |
" .dataframe tbody tr th {\n", |
|
|
74 |
" vertical-align: top;\n", |
|
|
75 |
" }\n", |
|
|
76 |
"\n", |
|
|
77 |
" .dataframe thead th {\n", |
|
|
78 |
" text-align: right;\n", |
|
|
79 |
" }\n", |
|
|
80 |
"</style>\n", |
|
|
81 |
"<table border=\"1\" class=\"dataframe\">\n", |
|
|
82 |
" <thead>\n", |
|
|
83 |
" <tr style=\"text-align: right;\">\n", |
|
|
84 |
" <th></th>\n", |
|
|
85 |
" <th>spicies</th>\n", |
|
|
86 |
" <th>ksent</th>\n", |
|
|
87 |
" </tr>\n", |
|
|
88 |
" </thead>\n", |
|
|
89 |
" <tbody>\n", |
|
|
90 |
" <tr>\n", |
|
|
91 |
" <th>0</th>\n", |
|
|
92 |
" <td>subtilis</td>\n", |
|
|
93 |
" <td>[-0.14380797743797302, 0.042256396263837814, 0...</td>\n", |
|
|
94 |
" </tr>\n", |
|
|
95 |
" <tr>\n", |
|
|
96 |
" <th>1</th>\n", |
|
|
97 |
" <td>subtilis</td>\n", |
|
|
98 |
" <td>[-0.05331391096115112, 0.06979767978191376, 0....</td>\n", |
|
|
99 |
" </tr>\n", |
|
|
100 |
" <tr>\n", |
|
|
101 |
" <th>2</th>\n", |
|
|
102 |
" <td>velezensis</td>\n", |
|
|
103 |
" <td>[-0.0810142382979393, 0.0927421897649765, 0.15...</td>\n", |
|
|
104 |
" </tr>\n", |
|
|
105 |
" <tr>\n", |
|
|
106 |
" <th>3</th>\n", |
|
|
107 |
" <td>velezensis</td>\n", |
|
|
108 |
" <td>[-0.11013756692409515, 0.08658633381128311, 0....</td>\n", |
|
|
109 |
" </tr>\n", |
|
|
110 |
" <tr>\n", |
|
|
111 |
" <th>4</th>\n", |
|
|
112 |
" <td>anthracis</td>\n", |
|
|
113 |
" <td>[-0.01958618313074112, 0.11950772255659103, 0....</td>\n", |
|
|
114 |
" </tr>\n", |
|
|
115 |
" </tbody>\n", |
|
|
116 |
"</table>\n", |
|
|
117 |
"</div>" |
|
|
118 |
], |
|
|
119 |
"text/plain": [ |
|
|
120 |
" spicies ksent\n", |
|
|
121 |
"0 subtilis [-0.14380797743797302, 0.042256396263837814, 0...\n", |
|
|
122 |
"1 subtilis [-0.05331391096115112, 0.06979767978191376, 0....\n", |
|
|
123 |
"2 velezensis [-0.0810142382979393, 0.0927421897649765, 0.15...\n", |
|
|
124 |
"3 velezensis [-0.11013756692409515, 0.08658633381128311, 0....\n", |
|
|
125 |
"4 anthracis [-0.01958618313074112, 0.11950772255659103, 0...." |
|
|
126 |
] |
|
|
127 |
}, |
|
|
128 |
"execution_count": 2, |
|
|
129 |
"metadata": {}, |
|
|
130 |
"output_type": "execute_result" |
|
|
131 |
} |
|
|
132 |
], |
|
|
133 |
"source": [ |
|
|
134 |
"df.head()" |
|
|
135 |
] |
|
|
136 |
}, |
|
|
137 |
{ |
|
|
138 |
"cell_type": "code", |
|
|
139 |
"execution_count": 19, |
|
|
140 |
"metadata": {}, |
|
|
141 |
"outputs": [ |
|
|
142 |
{ |
|
|
143 |
"data": { |
|
|
144 |
"text/plain": [ |
|
|
145 |
"(690000, 2)" |
|
|
146 |
] |
|
|
147 |
}, |
|
|
148 |
"execution_count": 19, |
|
|
149 |
"metadata": {}, |
|
|
150 |
"output_type": "execute_result" |
|
|
151 |
} |
|
|
152 |
], |
|
|
153 |
"source": [ |
|
|
154 |
"df.shape" |
|
|
155 |
] |
|
|
156 |
}, |
|
|
157 |
{ |
|
|
158 |
"cell_type": "code", |
|
|
159 |
"execution_count": 18, |
|
|
160 |
"metadata": {}, |
|
|
161 |
"outputs": [ |
|
|
162 |
{ |
|
|
163 |
"data": { |
|
|
164 |
"text/plain": [ |
|
|
165 |
"subtilis 321000\n", |
|
|
166 |
"velezensis 216000\n", |
|
|
167 |
"anthracis 153000\n", |
|
|
168 |
"Name: spicies, dtype: int64" |
|
|
169 |
] |
|
|
170 |
}, |
|
|
171 |
"execution_count": 18, |
|
|
172 |
"metadata": {}, |
|
|
173 |
"output_type": "execute_result" |
|
|
174 |
} |
|
|
175 |
], |
|
|
176 |
"source": [ |
|
|
177 |
"df.spicies.value_counts()" |
|
|
178 |
] |
|
|
179 |
}, |
|
|
180 |
{ |
|
|
181 |
"cell_type": "markdown", |
|
|
182 |
"metadata": {}, |
|
|
183 |
"source": [ |
|
|
184 |
"### Databunch" |
|
|
185 |
] |
|
|
186 |
}, |
|
|
187 |
{ |
|
|
188 |
"cell_type": "code", |
|
|
189 |
"execution_count": 3, |
|
|
190 |
"metadata": {}, |
|
|
191 |
"outputs": [], |
|
|
192 |
"source": [ |
|
|
193 |
"valid_idx = random.sample(range(df.shape[0]), int(np.floor(df.shape[0]* 0.2)))\n", |
|
|
194 |
"\n", |
|
|
195 |
"db = (ItemList.from_df(df,cols=\"ksent\").\n", |
|
|
196 |
" split_by_idx(valid_idx).\n", |
|
|
197 |
" label_from_df(cols=\"spicies\").\n", |
|
|
198 |
" databunch())" |
|
|
199 |
] |
|
|
200 |
}, |
|
|
201 |
{ |
|
|
202 |
"cell_type": "markdown", |
|
|
203 |
"metadata": {}, |
|
|
204 |
"source": [ |
|
|
205 |
"### Model" |
|
|
206 |
] |
|
|
207 |
}, |
|
|
208 |
{ |
|
|
209 |
"cell_type": "code", |
|
|
210 |
"execution_count": 4, |
|
|
211 |
"metadata": {}, |
|
|
212 |
"outputs": [], |
|
|
213 |
"source": [ |
|
|
214 |
"def submodel(dims, bias=False):\n", |
|
|
215 |
" layer_dims = list(zip(dims[:-1],dims[1:]))\n", |
|
|
216 |
" fcl = [nn.Linear(*x, bias=bias) for x in layer_dims]\n", |
|
|
217 |
" [nn.init.xavier_uniform_(m.weight) for m in fcl]\n", |
|
|
218 |
" if bias: \n", |
|
|
219 |
" for l in fcl: l.bias.data.normal_(0, 1)\n", |
|
|
220 |
" relu = [nn.ReLU() for _ in range(len(fcl))]\n", |
|
|
221 |
" layers = np.asarray(list(zip(fcl, relu))).ravel()[:-1]\n", |
|
|
222 |
" return nn.Sequential(*layers)\n" |
|
|
223 |
] |
|
|
224 |
}, |
|
|
225 |
{ |
|
|
226 |
"cell_type": "code", |
|
|
227 |
"execution_count": 5, |
|
|
228 |
"metadata": {}, |
|
|
229 |
"outputs": [], |
|
|
230 |
"source": [ |
|
|
231 |
"class Classifier (nn.Module):\n", |
|
|
232 |
"\n", |
|
|
233 |
" def __init__(self, encoder_dims, classifier_dims):\n", |
|
|
234 |
" super().__init__()\n", |
|
|
235 |
" self.encoder = submodel(encoder_dims,bias=True)\n", |
|
|
236 |
" self.classifier = submodel(classifier_dims,bias=True)\n", |
|
|
237 |
"\n", |
|
|
238 |
"\n", |
|
|
239 |
" def forward(self, x):\n", |
|
|
240 |
" x = self.encoder(x)\n", |
|
|
241 |
" return F.softmax(self.classifier(x), dim=1)\n", |
|
|
242 |
"\n", |
|
|
243 |
" def save_encoder(self,file:PathOrStr):\n", |
|
|
244 |
" torch.save(self.encoder.state_dict(), path)\n" |
|
|
245 |
] |
|
|
246 |
}, |
|
|
247 |
{ |
|
|
248 |
"cell_type": "code", |
|
|
249 |
"execution_count": 8, |
|
|
250 |
"metadata": { |
|
|
251 |
"scrolled": true |
|
|
252 |
}, |
|
|
253 |
"outputs": [], |
|
|
254 |
"source": [ |
|
|
255 |
"model = Classifier([100,50,3], [3,20,3]).double()" |
|
|
256 |
] |
|
|
257 |
}, |
|
|
258 |
{ |
|
|
259 |
"cell_type": "code", |
|
|
260 |
"execution_count": 9, |
|
|
261 |
"metadata": {}, |
|
|
262 |
"outputs": [ |
|
|
263 |
{ |
|
|
264 |
"data": { |
|
|
265 |
"text/plain": [ |
|
|
266 |
"Classifier(\n", |
|
|
267 |
" (encoder): Sequential(\n", |
|
|
268 |
" (0): Linear(in_features=100, out_features=50, bias=True)\n", |
|
|
269 |
" (1): ReLU()\n", |
|
|
270 |
" (2): Linear(in_features=50, out_features=3, bias=True)\n", |
|
|
271 |
" )\n", |
|
|
272 |
" (classifier): Sequential(\n", |
|
|
273 |
" (0): Linear(in_features=3, out_features=20, bias=True)\n", |
|
|
274 |
" (1): ReLU()\n", |
|
|
275 |
" (2): Linear(in_features=20, out_features=3, bias=True)\n", |
|
|
276 |
" )\n", |
|
|
277 |
")" |
|
|
278 |
] |
|
|
279 |
}, |
|
|
280 |
"execution_count": 9, |
|
|
281 |
"metadata": {}, |
|
|
282 |
"output_type": "execute_result" |
|
|
283 |
} |
|
|
284 |
], |
|
|
285 |
"source": [ |
|
|
286 |
"model" |
|
|
287 |
] |
|
|
288 |
}, |
|
|
289 |
{ |
|
|
290 |
"cell_type": "markdown", |
|
|
291 |
"metadata": {}, |
|
|
292 |
"source": [ |
|
|
293 |
"### Learner" |
|
|
294 |
] |
|
|
295 |
}, |
|
|
296 |
{ |
|
|
297 |
"cell_type": "code", |
|
|
298 |
"execution_count": 12, |
|
|
299 |
"metadata": {}, |
|
|
300 |
"outputs": [ |
|
|
301 |
{ |
|
|
302 |
"data": { |
|
|
303 |
"text/plain": [ |
|
|
304 |
"FlattenedLoss of CrossEntropyLoss()" |
|
|
305 |
] |
|
|
306 |
}, |
|
|
307 |
"execution_count": 12, |
|
|
308 |
"metadata": {}, |
|
|
309 |
"output_type": "execute_result" |
|
|
310 |
} |
|
|
311 |
], |
|
|
312 |
"source": [ |
|
|
313 |
"learn = Learner(db, model,metrics=[accuracy])\n", |
|
|
314 |
"learn.loss_func" |
|
|
315 |
] |
|
|
316 |
}, |
|
|
317 |
{ |
|
|
318 |
"cell_type": "code", |
|
|
319 |
"execution_count": 13, |
|
|
320 |
"metadata": { |
|
|
321 |
"scrolled": true |
|
|
322 |
}, |
|
|
323 |
"outputs": [ |
|
|
324 |
{ |
|
|
325 |
"data": { |
|
|
326 |
"text/html": [], |
|
|
327 |
"text/plain": [ |
|
|
328 |
"<IPython.core.display.HTML object>" |
|
|
329 |
] |
|
|
330 |
}, |
|
|
331 |
"metadata": {}, |
|
|
332 |
"output_type": "display_data" |
|
|
333 |
}, |
|
|
334 |
{ |
|
|
335 |
"name": "stdout", |
|
|
336 |
"output_type": "stream", |
|
|
337 |
"text": [ |
|
|
338 |
"LR Finder is complete, type {learner_name}.recorder.plot() to see the graph.\n" |
|
|
339 |
] |
|
|
340 |
}, |
|
|
341 |
{ |
|
|
342 |
"data": { |
|
|
343 |
"image/png": "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\n", |
|
|
344 |
"text/plain": [ |
|
|
345 |
"<Figure size 432x288 with 1 Axes>" |
|
|
346 |
] |
|
|
347 |
}, |
|
|
348 |
"metadata": { |
|
|
349 |
"needs_background": "light" |
|
|
350 |
}, |
|
|
351 |
"output_type": "display_data" |
|
|
352 |
} |
|
|
353 |
], |
|
|
354 |
"source": [ |
|
|
355 |
"learn.lr_find()\n", |
|
|
356 |
"learn.recorder.plot()" |
|
|
357 |
] |
|
|
358 |
}, |
|
|
359 |
{ |
|
|
360 |
"cell_type": "code", |
|
|
361 |
"execution_count": 14, |
|
|
362 |
"metadata": { |
|
|
363 |
"scrolled": true |
|
|
364 |
}, |
|
|
365 |
"outputs": [ |
|
|
366 |
{ |
|
|
367 |
"data": { |
|
|
368 |
"text/html": [ |
|
|
369 |
"<table border=\"1\" class=\"dataframe\">\n", |
|
|
370 |
" <thead>\n", |
|
|
371 |
" <tr style=\"text-align: left;\">\n", |
|
|
372 |
" <th>epoch</th>\n", |
|
|
373 |
" <th>train_loss</th>\n", |
|
|
374 |
" <th>valid_loss</th>\n", |
|
|
375 |
" <th>accuracy</th>\n", |
|
|
376 |
" <th>time</th>\n", |
|
|
377 |
" </tr>\n", |
|
|
378 |
" </thead>\n", |
|
|
379 |
" <tbody>\n", |
|
|
380 |
" <tr>\n", |
|
|
381 |
" <td>0</td>\n", |
|
|
382 |
" <td>0.588498</td>\n", |
|
|
383 |
" <td>0.583153</td>\n", |
|
|
384 |
" <td>0.968529</td>\n", |
|
|
385 |
" <td>00:35</td>\n", |
|
|
386 |
" </tr>\n", |
|
|
387 |
" <tr>\n", |
|
|
388 |
" <td>1</td>\n", |
|
|
389 |
" <td>0.593730</td>\n", |
|
|
390 |
" <td>0.606674</td>\n", |
|
|
391 |
" <td>0.943275</td>\n", |
|
|
392 |
" <td>00:35</td>\n", |
|
|
393 |
" </tr>\n", |
|
|
394 |
" <tr>\n", |
|
|
395 |
" <td>2</td>\n", |
|
|
396 |
" <td>0.594737</td>\n", |
|
|
397 |
" <td>0.589081</td>\n", |
|
|
398 |
" <td>0.962116</td>\n", |
|
|
399 |
" <td>00:35</td>\n", |
|
|
400 |
" </tr>\n", |
|
|
401 |
" <tr>\n", |
|
|
402 |
" <td>3</td>\n", |
|
|
403 |
" <td>0.600526</td>\n", |
|
|
404 |
" <td>0.583841</td>\n", |
|
|
405 |
" <td>0.967457</td>\n", |
|
|
406 |
" <td>00:36</td>\n", |
|
|
407 |
" </tr>\n", |
|
|
408 |
" <tr>\n", |
|
|
409 |
" <td>4</td>\n", |
|
|
410 |
" <td>0.591394</td>\n", |
|
|
411 |
" <td>0.605085</td>\n", |
|
|
412 |
" <td>0.945500</td>\n", |
|
|
413 |
" <td>00:35</td>\n", |
|
|
414 |
" </tr>\n", |
|
|
415 |
" <tr>\n", |
|
|
416 |
" <td>5</td>\n", |
|
|
417 |
" <td>0.586290</td>\n", |
|
|
418 |
" <td>0.582825</td>\n", |
|
|
419 |
" <td>0.968377</td>\n", |
|
|
420 |
" <td>00:36</td>\n", |
|
|
421 |
" </tr>\n", |
|
|
422 |
" <tr>\n", |
|
|
423 |
" <td>6</td>\n", |
|
|
424 |
" <td>0.584365</td>\n", |
|
|
425 |
" <td>0.580657</td>\n", |
|
|
426 |
" <td>0.970710</td>\n", |
|
|
427 |
" <td>00:36</td>\n", |
|
|
428 |
" </tr>\n", |
|
|
429 |
" <tr>\n", |
|
|
430 |
" <td>7</td>\n", |
|
|
431 |
" <td>0.586121</td>\n", |
|
|
432 |
" <td>0.583098</td>\n", |
|
|
433 |
" <td>0.968109</td>\n", |
|
|
434 |
" <td>00:35</td>\n", |
|
|
435 |
" </tr>\n", |
|
|
436 |
" <tr>\n", |
|
|
437 |
" <td>8</td>\n", |
|
|
438 |
" <td>0.580480</td>\n", |
|
|
439 |
" <td>0.580342</td>\n", |
|
|
440 |
" <td>0.970928</td>\n", |
|
|
441 |
" <td>00:36</td>\n", |
|
|
442 |
" </tr>\n", |
|
|
443 |
" <tr>\n", |
|
|
444 |
" <td>9</td>\n", |
|
|
445 |
" <td>0.578384</td>\n", |
|
|
446 |
" <td>0.579312</td>\n", |
|
|
447 |
" <td>0.972043</td>\n", |
|
|
448 |
" <td>00:37</td>\n", |
|
|
449 |
" </tr>\n", |
|
|
450 |
" </tbody>\n", |
|
|
451 |
"</table>" |
|
|
452 |
], |
|
|
453 |
"text/plain": [ |
|
|
454 |
"<IPython.core.display.HTML object>" |
|
|
455 |
] |
|
|
456 |
}, |
|
|
457 |
"metadata": {}, |
|
|
458 |
"output_type": "display_data" |
|
|
459 |
} |
|
|
460 |
], |
|
|
461 |
"source": [ |
|
|
462 |
"learn.fit_one_cycle(10,1e-2)" |
|
|
463 |
] |
|
|
464 |
}, |
|
|
465 |
{ |
|
|
466 |
"cell_type": "code", |
|
|
467 |
"execution_count": 16, |
|
|
468 |
"metadata": { |
|
|
469 |
"scrolled": true |
|
|
470 |
}, |
|
|
471 |
"outputs": [ |
|
|
472 |
{ |
|
|
473 |
"data": { |
|
|
474 |
"image/png": "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\n", |
|
|
475 |
"text/plain": [ |
|
|
476 |
"<Figure size 432x288 with 1 Axes>" |
|
|
477 |
] |
|
|
478 |
}, |
|
|
479 |
"metadata": { |
|
|
480 |
"needs_background": "light" |
|
|
481 |
}, |
|
|
482 |
"output_type": "display_data" |
|
|
483 |
} |
|
|
484 |
], |
|
|
485 |
"source": [ |
|
|
486 |
"learn.recorder.plot_metrics()" |
|
|
487 |
] |
|
|
488 |
}, |
|
|
489 |
{ |
|
|
490 |
"cell_type": "code", |
|
|
491 |
"execution_count": 17, |
|
|
492 |
"metadata": {}, |
|
|
493 |
"outputs": [], |
|
|
494 |
"source": [ |
|
|
495 |
"interpretation = learn.interpret()" |
|
|
496 |
] |
|
|
497 |
}, |
|
|
498 |
{ |
|
|
499 |
"cell_type": "code", |
|
|
500 |
"execution_count": 30, |
|
|
501 |
"metadata": {}, |
|
|
502 |
"outputs": [ |
|
|
503 |
{ |
|
|
504 |
"data": { |
|
|
505 |
"image/png": "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\n", |
|
|
506 |
"text/plain": [ |
|
|
507 |
"<Figure size 432x288 with 1 Axes>" |
|
|
508 |
] |
|
|
509 |
}, |
|
|
510 |
"metadata": { |
|
|
511 |
"needs_background": "light" |
|
|
512 |
}, |
|
|
513 |
"output_type": "display_data" |
|
|
514 |
} |
|
|
515 |
], |
|
|
516 |
"source": [ |
|
|
517 |
"interpretation.plot_confusion_matrix()" |
|
|
518 |
] |
|
|
519 |
}, |
|
|
520 |
{ |
|
|
521 |
"cell_type": "code", |
|
|
522 |
"execution_count": null, |
|
|
523 |
"metadata": {}, |
|
|
524 |
"outputs": [], |
|
|
525 |
"source": [] |
|
|
526 |
} |
|
|
527 |
], |
|
|
528 |
"metadata": { |
|
|
529 |
"kernelspec": { |
|
|
530 |
"display_name": "Python 3", |
|
|
531 |
"language": "python", |
|
|
532 |
"name": "python3" |
|
|
533 |
}, |
|
|
534 |
"language_info": { |
|
|
535 |
"codemirror_mode": { |
|
|
536 |
"name": "ipython", |
|
|
537 |
"version": 3 |
|
|
538 |
}, |
|
|
539 |
"file_extension": ".py", |
|
|
540 |
"mimetype": "text/x-python", |
|
|
541 |
"name": "python", |
|
|
542 |
"nbconvert_exporter": "python", |
|
|
543 |
"pygments_lexer": "ipython3", |
|
|
544 |
"version": "3.6.9" |
|
|
545 |
}, |
|
|
546 |
"latex_envs": { |
|
|
547 |
"LaTeX_envs_menu_present": true, |
|
|
548 |
"autoclose": false, |
|
|
549 |
"autocomplete": true, |
|
|
550 |
"bibliofile": "biblio.bib", |
|
|
551 |
"cite_by": "apalike", |
|
|
552 |
"current_citInitial": 1, |
|
|
553 |
"eqLabelWithNumbers": true, |
|
|
554 |
"eqNumInitial": 1, |
|
|
555 |
"hotkeys": { |
|
|
556 |
"equation": "Ctrl-E", |
|
|
557 |
"itemize": "Ctrl-I" |
|
|
558 |
}, |
|
|
559 |
"labels_anchors": false, |
|
|
560 |
"latex_user_defs": false, |
|
|
561 |
"report_style_numbering": false, |
|
|
562 |
"user_envs_cfg": false |
|
|
563 |
}, |
|
|
564 |
"notify_time": "30", |
|
|
565 |
"toc": { |
|
|
566 |
"base_numbering": 1, |
|
|
567 |
"nav_menu": {}, |
|
|
568 |
"number_sections": true, |
|
|
569 |
"sideBar": true, |
|
|
570 |
"skip_h1_title": false, |
|
|
571 |
"title_cell": "Table of Contents", |
|
|
572 |
"title_sidebar": "Contents", |
|
|
573 |
"toc_cell": false, |
|
|
574 |
"toc_position": {}, |
|
|
575 |
"toc_section_display": true, |
|
|
576 |
"toc_window_display": false |
|
|
577 |
} |
|
|
578 |
}, |
|
|
579 |
"nbformat": 4, |
|
|
580 |
"nbformat_minor": 2 |
|
|
581 |
} |